Reproductive tract microbiota of women in childbearing age shifts upon type of gynecological infections and phase in physiological cycle

Background: The association of the normal physiological cycle to the structural pattern of microbiota in reproductive tract of women at reproductive age has not been extensively explored. This study was undertaken to determine whether the vaginal microbes of women at childbearing age is different among groups dened by urogenital tract infections, childbearing history and menstrual cycle, respectively. Results: This was a multiple case-control study of women at childbearing age who were assigned to case or control groups according to their states of urogenital tract infections. The participants were also grouped by childbearing history and menstrual cycle. Samples of vaginal swabs were collected and stored at -70 ℃ until assayed. The V3-V4 regions of 16S rRNA genes were amplied using PCR and sequenced on the Illumina MiSeq platform. We tested the hypothesis of whether the relative abundance of microbial species in vaginal microbiota was different between women with different urogenital tract infections, childbearing history and menstrual cycle. We showed that the vaginal microbial richness(Alpha diversity measured by PD_whole tree) was decreased in normal women(without reproductive tract infections) than in those with bacterial vaginosis (BV), and decreased in pregnant women than in other groups of non-pregnancy. Similarly, women from groups of normal and in pregnancy had lower beta diversity on measure of unweighted_unifrac distance in comparison to those of uninfected and non-pregnant. The top 10 genus relative abundance, especially that Lactobacillus was the most dominant genus with the relative abundance of 71.55% among all samples, did not differ signicantly between groups of childbearing history and menstrual cycle analyzed by ANOVA and nonparametric kruskal_wallis.

stored at -70℃ until assayed. The V3-V4 regions of 16S rRNA genes were ampli ed using PCR and sequenced on the Illumina MiSeq platform. We tested the hypothesis of whether the relative abundance of microbial species in vaginal microbiota was different between women with different urogenital tract infections, childbearing history and menstrual cycle. We showed that the vaginal microbial richness(Alpha diversity measured by PD_whole tree) was decreased in normal women(without reproductive tract infections) than in those with bacterial vaginosis (BV), and decreased in pregnant women than in other groups of non-pregnancy. Similarly, women from groups of normal and in pregnancy had lower beta diversity on measure of unweighted_unifrac distance in comparison to those of uninfected and non-pregnant. The top 10 genus relative abundance, especially that Lactobacillus was the most dominant genus with the relative abundance of 71.55% among all samples, did not differ signi cantly between groups of childbearing history and menstrual cycle analyzed by ANOVA and nonparametric kruskal_wallis.
Lactobacillus iners and Lactobacillus helveticus have the most abundance, totally account for 97.92% relative abundance of genus Lactobacillus. It is proposed that a higher L.helveticus/L.iners ratio is more likely to present in normal women than in the infected and in pregnant than in non-pregnant, although this comparison lacks statistical signi cance.
Conclusions: The relative abundance of dominant bacterial taxa in vaginal microbial communities of women at childbearing age, characterized with 16S rRNA gene sequence and QIIME based analysis, were not different among groups of childbearing history and menstrual cycle. Women from groups of in pregnancy and without reproductive tract infections had lower alpha and beta diversity. The compositional ratio of the main lactobacillus species may shift depending on the normal physiological cycle and reproductive tract infections.

Background
Woman genital tract is one of the natural niches that harbor the most amount of body microbes [1].The normal structure of vaginal microbiome plays a pivotal role in maintaining the healthy microecosystem, especially for those in childbearing age. Any deviation of this eubiosis can lead to disorders such as bacterial vaginosis(BV), mycotic vaginitis(MV) or pelvic in ammation, which signi cantly impact the health of women, their fetuses and new born infants [2][3][4]. Furthermore, the role of dysbiosis in causing gynecological cancers has been appreciated only recently [5]. On the other hand, the composition of genital tract microbiota differs depending on factors that have no direct relation to infection, such as race, nationality that the women belonging to [6], or hygiene practices that they accustomed to [7,8]. Quite few reports dealt with the impact of a natural menstrual cycle on bacterial growth, colonization, and community structure. The others were about the comparison of vaginal microbial composition of nonpregnant with pregnant women by the case-control longitudinal study but these designs did not re ect on the real menstrual cycle [9,10]. We thus postulate that gynecological infections as well as the normal physiological cycle may affect the structural pro le of vaginal microbiota of women in childbearing age.
So the current study sought to examine this question by characterizing the differential microbiota among groups of urogenital tract infections, childbearing history and phases of menstrual cycle.

Results
The grouping and characteristics of the 111 study subjects were shown in Additional le: Table S1 and were found by childbearing history, menstrual cycle, gynecological infections, respectively. The mean age of women without childbearing was signi cantly lower than that of women with childbearing and women in pregnancy.
The alpha diversity on measure of PD_whole tree other than chao1, observed_outs and Shanonn's index was lower in group of women in pregnancy compared to those with_childbearing(P=0.021) and those in follicular phase(P=0.048). Of the 5 groups on gynecological diagnosis, normal women had lower alpha diversity on measure of PD_whole tree in contrast to the other infected groups, but only the difference to BV group was statistically signi cant(P=0.04) (Fig 1a-c, Additional le: Table S2). As for the measurement of beta diversity, normal women were clustered together based on their genital tract microbial taxa having closer genetic a nity(having shorter unweighted_unifrac distance) apart from those in women of BV, mycotic vaginitis(MV) and MV+BV group(Bonferroni-corrected parametric P 0.001, Bonferroni-corrected nonparametric P=0.136) (Fig 2a, Additional le: Table S3). With respect to grouping by childbearing history and menstrual cycle, women in pregnancy had conspicuous constitutional similarity of microbial population compared to those with or without childbearing group(Bonferronicorrected parametric P=0.001, Bonferroni-corrected nonparametric P=0.028.) (Fig 2b, Additional le: Table  S3) Similarly, taxa in genital tract of pregnant women were signi cantly alienated from taxa in women that were in luteal_phase or follicular_phase(Bonferroni-corrected parametric P 0.001, Bonferronicorrected nonparametric P=0.028). (Fig 2c, Additional le: Table S3).
The top 10 genus relative abundance among different groups did not differ signi cantly compared by ANOVA and nonparametric kruskal_wallis, with one exception that groups de ned by gynecological infections had different relative abundance of 3 genera out of the top 10 genera tested by ANOVA (Fig 1d,  Fig.3, Additional le: Table S4). The detail of these differences were intensively analysed by Bonferroni multiple comparisons. That is, group of mycotic vaginitis(MV) had higher relative abundance of genus Gardnerella compared to group of normal(P=0.041) and had higher relative abundance of genus NA(a liated to Lactobacillales orders) compared to group of normal(P=0.042) and group of BV(P=0.025) (Fig 1d, Additional le: Table S5). Women affected nongonococcal urethritis/cervicitis(NUC) had higher relative abundance of Streptococcus in contrast to the others(P<0.001) (Additional le: Table S5). The graph of NUC group was not included in Figure 1d due to its 95% interval line disproportionately expanded caused by the low sample size(only 2 women). Lactobacillus is the most dominant over other taxa for sustaining normal vaginal ecosystem. As shown in gure 1d and gure 3, Lactobacillus was the most dominant genus among all sample groups with the mean relative abundance of Lactobacillus being the highest in Normal group over other infection groups. But there were no signi cant relative abundance discrepancies between groups of gynecological infection in genus Lactobacillus, nor were in Atopobium and Prevotella, even though which had been classi ed as bacterial vaginosis associated bacteria(BVAB) or as markers for the other forms of urogenital tract infections(data not shown).
In addition to the common sense that Lactobacillus is the dominant population in normal (without infectious diseases) group of women, we found women in gestation also had a slight higher Lactobacillus relative abundance in comparison to women without pregnancy. Furthermore, women in gestation and without genital tract infections (in_pregnancy+normal) harbored the highest proportion of genus Lactobacillus over the other group of women (Fig 3). In order to scrutinize the exquisite constituent structure of genus Lactobacillus, we explored the compositional pro le of all the 11 Lactobacillus species wherein Lactobacillus iners and Lactobacillus helveticus have the most abundance, totally account for 97.92% relative abundance of genus Lactobacillus. It is proposed that a higher helveticus/iners ratio is more likely to present in normal women than in the infected and in pregnant than in non pregnant, although this comparison lacks statistical signi cance(at 0.05 signi cant level) (Fig 4) .

Discussion
We established that the structural pattern of vaginal microbiota characterized by high relative abundance of species of Atopobium as well as the presence of Prevotella, Sneathia, Gardnerella, Ruminococcaceae, Parvimonas, Mobiluncus and other taxa previously shown to be bacterial vaginosis associated bacteria(BVAB) was rarely observed in either pregnant or other groups of women. Nevertheless, it is found in all the samples that only Gardnerella and Streptococcus were signi cant indicator genus for mycotic vaginitis(MV) and nongonococcal urethritis/cervicitis(NUC), respectively. Although there were no signi cant relative abundance differences of the top 10 bacterial genus, which totally account for 96.25% microbe amount of the whole community, between pregnant and non-pregnant women, we observed the decreased microbial richness measured by PD_whole tree alpha diversity in pregnant than in non-pregnant women and in normal than in infected women. This result are contrary to recently reported by Yulian Chen,et al [11] in that they found both pregnancy and HPV infection can increase vaginal bacterial microbial richness and diversity. Although we consider the reason is they measured the microbial richness by Chao and Shannon index other than we used PD_whole tree that focus on the phylogenic a nity between each microbe, there must exist underlying factors that shield the real association of taxonomic diversity to pregnancy. Some intensive studies involving metagenome or 16S RNA gene sequencing are needed to identify the genomic function differences underlying the taxa diversity differences between microbiota in the corresponding groups de ned by urogenital infections or normal physiological cycle [12][13][14].
Lactobacillus was the most dominant genus with the relative abundance of 71.55% among all 111 samples in this study. However, we did not nd the presence of Lactobacillus vaginalis, L. crispatus, L.gasseri and L. jensenii at species level as reported in previous studies [15][16][17]. Indeed, the dominant members of Lactobacillus species has been limited and elusive based on the context of each study. In most cases, the vaginal microbiota through a menstrual cycle demonstrate that L.crispatus, L.iners, and L.jensenii was the dominant members [9]. Lactobacillus gasseri was considered prominent in other studies [15,16], but it was not a dominant organism in any woman's microbiota pro le other than L.iners, and L. helveticus in this study, nor were L.crispatus, L. vaginalis, and L.jensenii.
There has been recent species-speci c attention to L.iners which is commonly found in the vagina [15] and has been associated with both BV and healthy states [18][19][20][21]. In addition, L.iners is often the rst Lactobacillus species to recover after treatment for BV [21,22]. But the role of L.iners in the dynamic balance between the family members of vaginal microbiota is still controversial. Some reports suggested there are strains of L.iners which are highly stable over time while others are associated with a rapidly changing vaginal microbiota toward a BV state [16,22]. Jakobsson et al [23] reported that L iners is a dominant part of the vaginal ora when the ora is in a transitional stage from normal to abnormal while others observed that L.crispatus produce more H 2 O 2 and bactericidal agents compared with other Lactobacilli species, so that it is considered the leading role in maintaining the health of vaginal ecosystem [24]. Unfortunately, we did not found L.crispatus in the specimens in the present and our previous study samples [25]. According to our ndings, L.iners and L.helveticus totally account for 97.92% relative abundance of genus Lactobacillus with L.helveticus was discovered for the rst time in this study as a leading member in genus Lactobacillus. Moreover, the increased proportion of L.helveticus over L.iners in normal and pregnant women than in other group of women indicates it may be the indicator for healthy status of reproductive tract microecology. We thus conjecture that women with L.iners-dominant and L.crispatus-absent intravaginal microbial ora, who is presumed to be normal by the phenotype methods, is thought to be at risk of BV or other urogenital tract infections and must be monitored carefully. Finished and ongoing work is needed to evaluate the genomic heterogeneity of L.iners and if different strains are associated with health and abnormal outcomes [26].
In brief, the current study engaged on multiple designs to explore the compositional differences of vaginal microbes between groups of women at reproductive age de ned by urogenital tract infections, childbearing history and menstrual cycle, respectively. The vast volume work of sequencing data processing and analysis was performed on QIIME2 and QIIME 1.9.0 platform embedded in Linux operational system in addition to some outcomes resulted from QIIME was analyzed by SPSS 22.0. These methods should have identi ed signi cant differences among groups of women de ned by urogenital tract infections, childbearing history and menstrual cycle. Weaknesses include the modest sample size, especially for NUC sample group that had only 2 women, which contributed to the larger variance in the comparison of outcomes between groups using parametric methods such as ANOVA. The study only allows us to detect the effect size of different sample grouping schemes without the control of confounding factors such as age, occupation, and other personal behavioral factors.

Conclusions
Our study showed that women without reproductive tract infections or in pregnancy had decreased vaginal microbial richness and beta diversity. The compositional ratio of the main lactobacillus species may shift depending on the normal physiological cycle and reproductive tract infections. We found no bacterial vaginosis associated bacteria in addition to Gardnerella. The species level pro le of Lactobacillus were also differed from other studies. We suspect this discrepancy was due to the different source of samples on whom there were diverse formulae of antibiotic administration, for instance, antimicrobials abuse is popular in present in China.

Methods
Study subjects 111women at reproductive age (32.61±7.75 years old) were recruited at Qilu Hospital of Shandong University and the 5th People's Hospital of Jinan city, China, and were grouped by reproductive tract infections (gynecological diagnosis), childbearing history and the phase of menstrual cycle on the day the specimen swabs were collected(Additional le: Figure S1).
Generally, the rst day of menstruation to the day before the next menstruation is called a menstrual cycle [27]. According to the average 28 days of each menstrual cycle, 111 women of childbearing age in this study were divided into 3 stages in proportion to the average menstrual cycle: (1) Follicular_phase is equivalent to the interval from the 5th to 14th day of a menstrual cycle.
(2) Luteal_phase is the period from the beginning of ovulation to the next menstruation, i.e. the 15th to 28th day of a menstrual cycle.
(3) In_pregnancy was de ned as ≥ 40 days pregnancy and before parturition. Because women in menstrual period (equivalent to the period from the 1st to 4th day of a menstrual cycle) did not visit the hospital, there were no women of menstrual period recruited to this study.
The grouping of childbearing history was performed with respect to the results of questionnaire and clinical examination.
Nugent score and Amsel's criteria in combination with clinical symptoms were used to identify normal women from groups of different reproductive tract infections [28].
Individuals with antibiotic usage within the last 2 weeks or those with examination and treatment of urogenital tract within the last 3 days were excluded. The study protocol was approved by the study review board and ethic committee of Institute of Basic Medicine Shandong First Medical University & Shandong Academy of Medical Sciences. All the subjects signed informed consent and lled out questionnaires.

Biospecimen collection
We got two swabs from each subject with one dry swab thoroughly wiped the lateral and posterior fornix of the vaginal wall to collect the complete population of microbiome, while the other physiological saline soaked swab were used to extract vaginal discharge for measurement of pH and direct detection of mycotic infection under optical microscopy.
16S rRNA gene sequence analysis The DNA was extracted from the dry swab of each sample as described previously [25,29] Table S6). These single-end reads were processed for quality ltering, denoising, and chimera removal to form Amplicon Sequence Variants(ASVs, also known as Operational Taxonomy Units(OTUs)) using the DADA2 plugin [31] in QIIME2(Quantitative Insights into Microbial Ecology, version: 2019.10 (1572640561)) [32]. OTUs with only one read were excluded. Alpha diversity was estimated by the measure of observed_otus, Shannon's Index(using information of OTU frequency) [33], chao1 and PD_whole_tree(using information of phylogenetic relationship of different OTUs) [34] using a sampling depth range of 10-25,000 seqences/sample to generate 10 rare ed OTU tables and compute alpha diversity metrics for each rare ed OTU table. Beta diversity was measured by unweighted UniFrac, weighted UniFrac and bray_curtis distance based on the 25,000 seqences/sample rare ed OTU table [35]. Relative abundances of taxa summarized at the L2(Phylum) to L6(genus) levels were also respectively calculated from the 25,000 seqences/sample rare ed OTU table. These analyses were conducted using the work ow core_diversity_analyses.py in QIIME(version 1.9.1) [36] and Green genes data based classi er (version gg-13-8-99-nb)[37] as the taxonomic reference.

Statistical analysis
The tests of alpha diversity(PD_whole_tree, chao1, observed_otus and Shannon's Index) and beta diversity(unweighted UniFrac, weighted UniFrac and bray_curtis distance) differences between groups were performed using a two-sided Student's two-sample t-test embedded in the QIIME 1.9.1 work ow core_diversity_analyses.py. Taxa relative abundance differences over groups of gynecological diagnosis, childbearing history and menstrual cycle were estimated based on parametric ANOVA and nonparametric kruskal_wallis in QIIME 1.9.1 command group_signi cance.py. The signi cant relative abundance differences of 3 bacterial genera were intensively compared between each 2 groups by Post Hoc Bonferroni multiple comparisons in SPSS(version 22.0). Bonferroni correction was used to adjust for tests of multiple taxa. P values less than 0.05 were considered signi cant after adjustment for multiple tests.